eggd_dias_batch (DNAnexus Platform App)

What are typical use cases for this app?

DNAnexus app for launching CNV calling, one or more of SNV, CNV and mosaic reports workflows as well as eggd_artemis from a given directory of Dias single output and a manifest file(s).

---

What inputs are required for this app to run?

Required

• -iassay (str): string of assay to run analysis for (CEN or TWE), used for searching of config files automatically (if -iassay_config_file not specified)
• -isingle_output_dir (str): path to output directory of Dias single to use as input files
• -imanifest_files (array:file): one or more manifest files from Epic or Gemini, maps sample ID -> required test codes / HGNC IDs (required for running any reports mode)
• one or more running modes, see running modes

Useful ones

Strings

• -iassay_config_dir (str): DNANexus project:path to directory containing config files, the highest version for the given -iassay string will be used
• -icnv_call_job_id (str): job ID of cnv calling job to use for generating CNV reports if CNV calling is not first being run (n.b. this is mutually exclusive with -icnv_call)
• -iexclude_samples (str): comma separated string of samples to exclude from CNV calling / CNV reports (these should be formatted as InstrumentID-SpecimenID (i.e. 123245111-33202R00111))
• -imanifest_subset (str): comma separated string of Epic samples in manifest on which to ONLY run jobs (these should be formatted as InstrumentID-SpecimenID (i.e. 123245111-33202R00111)). This option is to be used if an Epic batch had mistakes in the manifest, which have now been corrected. This will filter the updated batch so that reports jobs are only run for the corrected samples and not the whole batch.

Files

• -iqc_file (file): xlsx file mapping QC state of each sample (this is an optional input file for eggd_artemis, and should only be specified with -iartemis=true)
• -imultiqc_report (file): HTML MultiQC report (this is an optional input file for eggd_artemis and should only be specified with -iartemis=true, it will pass the file as input to suppress searching for a MultiQC job in the project)
• -iassay_config_file (file): Config file for assay, if not provided will search default assay_config_dir for highest version config file for the given -assay string
• -iexclude_samples_file (file): file of samples to exclude from CNV calling / CNV reports, one sample name per line. Epic samples should be formatted as InstrumentID-SpecimenID (i.e. 123245111-33202R00111) Example formatting of exclude_samples_file`:

  X225201
  125558769-23272R0123
  

Booleans

• -iexclude_controls (bool): controls if to automatically exclude control samples from CNV calling based on the regex pattern '^\w+-\w+Q\w+-' (default: true)
• -isplit_tests (bool): controls if to split multiple panels / genes in a manifest to individual reports instead of being combined into one
• -iunarchive (bool): controls whether to automatically unarchive any required files that are archived. Default is to fail the app with a list of files required to unarchive. If set to true, all required files will start to be unarchived and the job will exit with a zero exit code and the job tagged to state no jobs were launched
• -iunarchive_only (bool): controls if to only run the app to check for archived files and unarchive (i.e no launching of jobs), if all files are found in an unarchived state the app will exit with a zero exit code.
  • n.b. in this mode, unarchive defaults to True and unarchiving will always be run

Running modes

• -icnv_call (bool): controls if to run CNV calling (n.b. this is mutually exclusive with -icnv_call_job_id)
• -icnv_reports (bool): controls if to run CNV reports workflows
• -isnv_reports (bool): controls if to run SNV reports workflows
• -imosaic_reports (bool): controls if to run mosaic reports workflow
• -iartemis (bool): controls if to run eggd_artemis

n.b. the default is for all running modes to be false, therefore if none are specified the app will raise and error and exit

Testing

• -itesting (bool): controls if to run in testing mode and terminate all launched jobs after launching
• -isample_limit (int): no. of samples to launch jobs for, used during testing to speed up running of app

---

How does this app work?

The app takes as a minimum input a path to Dias single output, an assay config, and at least one of the above listed running modes. The default behaviour is to pass an assay string specified to run for (with -iassay), which will search DNAnexus for the highest version config file in -iassay_config_dir (default: 001_Reference:/dynamic_files/dias_batch_configs/) and use this for analysis. Alternatively, an assay config file may be specified to use instead with -iassay_config_file. If running a reports workflow a manifest file must also be specified.

Before any jobs are launched, a check of the archival state of all required files is first made. This will use the file pattern mappings either defined in utils.defaults or from the assay config file (if specified) to search for the per sample and per run files required, any will raise an error on any archived files if unarchive=True is not set.

The general behaviour of each mode is as follows:

CNV calling

Minimum inputs:

• -iassay or -iassay_config_file
• -isingle_output_dir

Behaviour:

• Check if inputs provided are valid
• Search for and download latest config for assay (if not provided directly)
• Parse through config file to add reference files to input fields
• Download and format genepanels file
• Search for bam files using the folder and name provided in the config file under -isingle_output_dir
  • Remove any files belonging to samples specified to -iexclude_samples
• Run CNV calling app
  • n.b. if -icnv_reports=true is specified, the app will be held until CNV calling completes, and the output will be used for launching CNV reports
• Launch reports workflow (if any specified; see below)
• Write summary report and upload

Reports workflows

Minimum inputs:

• -iassay or -iassay_config_file
• -isingle_output_dir
• -imanifest_files
• -icnv_reports -> -icnv_call=true OR -icnv_call_job_id

Behaviour:

• Check if inputs provided are valid
• Search for and download latest config for assay (if not provided directly)
• Parse through config file to add reference files to input fields
• Download and format genepanels file
• Download manifest(s)
  • Check provided test codes are valid and present in genepanels file
  • Get full panel and clinical indication strings for each test code from genepanels file
• For CNV reports:
  • Gather all segments.vcf files from CNV call job output
  • Find excluded intervals bed file from CNV call job output
  • Find previous xlsx reports (used for setting report name suffix)
  • Filter manifest by samples having a VCF found
  • Read CNV reports inputs from config, parse in string inputs (i.e. panel str to workbooks) and add vcf input for VEP
  • Check for previous xlsx reports for same sample and increment to always be +1
  • Launch CNV reports workflow
• For SNV/mosaic reports:
  • Gather all VCF and mosdepth files from sub dir and name pattern specified in config as input to VEP and Athena, respectively
  • Find previous xlsx reports (used for setting report name suffix)
  • Filter manifest by samples having VCF and mosdepth files found
  • Read SNV/mosaic reports inputs from config, parse in string inputs (i.e. panel str to workbooks), add VCF input for VEP and mosdepth files for Athena
  • Check for previous xlsx reports for same sample and increment to always be +1
  • Launch SNV/mosaic reports workflow

n.b.

• if -itesting=true is specified, reports jobs will launch but not start running, and will be automatically terminated on the app completing

Artemis

Minimum inputs

• -isnv_reports and / or -icnv_reports
• -iqc_file

Behaviour

• check if inputs provided are valid
• check if one or more jobs launched for SNV / CNV reports
  • get parent path of both if true to set as input
• optionally check if QC xlsx provided to use as input
• launch eggd_artemis, will be dependent on all SNV and CNV report workflows completing

---

Example commands

Running CNV calling and CNV reports for CEN assay:

dx run app-eggd_dias_batch \
    -iassay=CEN \
    -imanifest_files=file-xxx \
    -isingle_output_dir=project-xxx:/path_to_output/ \
    -icnv_call=true \
    -icnv_reports=true

Running reports for CNV and SNV (using previous CNV calling output) and launching eggd_artemis:

dx run app-eggd_dias_batch \
    -iassay=CEN \
    -imanifest_files=file-xxx \
    -isingle_output_dir=project-xxx:/path_to_output/ \
    -icnv_call_job_id=job-xxx \
    -icnv_reports=true \
    -isnv_reports=true \
    -iartemis=true \
    -iqc_file=file-xxx

Running SNV reports with specified config file:

dx run app-eggd_dias_batch \
    -iassay_config_file=file-xxx \
    -imanifest_files=file-xxx \
    -isingle_output_dir=project-xxx:/path_to_output/ \
    -isnv_reports=true

Running all modes in testing:

dx run app-eggd_dias_batch \
    -iassay=CEN \
    -imanifest_files=file-xxx \
    -isingle_output_dir=project-xxx:/path_to_output/ \
    -icnv_call=true \
    -icnv_reports=true \
    -isnv_reports=true \
    -imosaic_reports=true

Running CNV calling, CNV reports, SNV reports and Artemis with 2 manifest files:

dx run app-eggd_dias_batch \
    -iassay=CEN \
    -isingle_output_dir=project-xxx:/path_to_output/ \
    -imanifest_files=file-xxx \
    -imanifest_files=file-yyy \
    -iqc_file=file-zzz \
    -icnv_call=true \
    -icnv_reports=true \
    -isnv_reports=true \
    -iartemis=true

---

Config file design

The config file for an assay is written in JSON format and specifies the majority of inputs for running each type of analysis. A populated example config file may be found here.

The top level section should be structured as follows:

{
    "assay": "CEN",
    "version": "2.2.0",
    "cnv_call_app_id": "app-GJZVB2840KK0kxX998QjgXF0",
    "snv_report_workflow_id": "workflow-GXzkfYj4QPQp9z4Jz4BF09y6",
    "cnv_report_workflow_id": "workflow-GXzvJq84XZB1fJk9fBfG88XJ",
    "reference_files": {
        "genepanels": "project-Fkb6Gkj433GVVvj73J7x8KbV:file-GVx0vkQ433Gvq63k1Kj4Y562",
        "exons_nirvana": "project-Fkb6Gkj433GVVvj73J7x8KbV:file-GF611Z8433Gk7gZ47gypK7ZZ",
        "genes2transcripts": "project-Fkb6Gkj433GVVvj73J7x8KbV:file-GV4P970433Gj6812zGVBZvB4",
        "exonsfile": "project-Fkb6Gkj433GVVvj73J7x8KbV:file-GF611Z8433Gf99pBPbJkV7bq"
    },
    "name_patterns": {
        "Epic": "^[\\d\\w]+-[\\d\\w]+",
        "Gemini": "^X[\\d]+"
    },
    ...

• assay (str) : assay type the config is for, used for finding highest version config file when -iassay is specified
• version (str) : the version of this config file
• {cnv_call_app|_report_workflow}_id (str) : the IDs of CNV calling and reports workflows to use
• reference_files (dict) : mapping of reference file name : DNAnexus file ID, reference file name must be given as shown above, and DNAnexus file ID should be provided as project-xxx:file-xxx
• name_patterns (dict) : mapping of the manifest source and a regex pattern to use for filtering sample names and files etc.
• mode_file_patterns (dict | optional): mapping for each running mode to sample and run file patterns for which to search and check the archival state of before launching any jobs. Defaults are defined in utils.defaults, and a mapping of the same structure may be added to the assay config file to override the defaults.

The definitions of inputs for CNV calling and each reports workflow should be defined under the key modes, containing a mapping of all inputs and other inputs for controlling running of analyses.

Example format of CNV call app structure:

"modes": {
    "cnv_call": {
        "instance_type": "mem2_ssd1_v2_x8",
        "inputs": {
            "bambais": {
                "folder": "/sentieon-dnaseq-4.2.1/",
                "name": ".bam$|.bam.bai$"
            },
            "GATK_docker": {
                "$dnanexus_link": {
                    "$dnanexus_link": {
                        "project": "project-Fkb6Gkj433GVVvj73J7x8KbV",
                        "id": "file-GBBP9JQ433GxV97xBpQkzYZx"
                    }
                }
            },
            "annotation_tsv": {
                ...

• instance_type (str; optional) : instance type to use when running CNV calling app
• inputs (dict) : mapping of each app input field to required input
  • bambais is a dynamic input and BAM files are parsed at run time using the folder and name keys, folder will be used as a sub folder under the -isingle_output_dir specified, and name will be used as a regex pattern for finding files
  • other inputs should be specified in the standard $dnanexus_link mapping format to be passed directly to the underlying run API call

Example format of a reports workflow structure:

"cnv_reports": {
        "stage_instance_types": {
            "stage-cnv_vep.vcf": "mem2_ssd2_v2_x72"
        },
        "inputs": {
            "stage-cnv_generate_bed_vep.exons_nirvana": "INPUT-exons_nirvana",
            "stage-cnv_generate_bed_vep.nirvana_genes2transcripts": "INPUT-genes2transcripts",
            "stage-cnv_generate_bed_vep.gene_panels": "INPUT-genepanels",
            "stage-cnv_generate_bed_vep.flank": 495,
            "stage-cnv_generate_bed_vep.additional_regions": {
                "$dnanexus_link": {
                    "project": "project-Fkb6Gkj433GVVvj73J7x8KbV",
                    "id": "file-GJZQvg0433GkyFZg13K6VV6p"
                }
            },
            "stage-cnv_vep.config_file": {
                "$dnanexus_link": {
                    "project": "project-Fkb6Gkj433GVVvj73J7x8KbV",
                    "id": "file-GQGJ3Z84xyx0jp1q65K1Q1jY"
                }
            },
            "stage-cnv_vep.vcf": {
                "folder": "CNV_vcfs",
                "name": "_segments.vcf$"
            },

• stage_instance_types (dict; optional) : mapping of stage-name to instance type to use, this will override the app and workflow defaults

• inputs (dict) : mapping of each stage input field to required input

  • inputs may be defined as regular integers / strings / booleans, $dnanexus_link file mappings or using INPUT- placeholders
  • INPUT- placeholders may be defined in the config to pass variable inputs such as files and strings at runtime as workflow inputs. See the placeholder section below for details.
  • inputs for the following stages follow the same behaviour as the bambais input for the CNV calling app of being provided as a "folder" and "name" key for searching:
    • cnv_reports:
      • stage-cnv_vep.vcf
    • snv_reports and mosaic_reports
      • stage-rpt_vep.vcf
      • stage-rpt_athena.mosdepth_files
  • example format of the above:

    # example of finding VCFs from Sentieon
    # this will find any vcf|vcf.gz but NOT .g.vcf
    "stage-rpt_vep.vcf" :{
        "folder": "sentieon-dnaseq",
        "name": "^[^\.]*(?!\.g)\.vcf(\.gz)?$"
    }
    

  Placeholders

  Placeholder strings may be defined in the config file for parsing in reference files and strings that are generated at run time. These are split between the reference files defined in the top level of the config (i.e. the genepanels file) or strings (such as the panel name).

  For reference files, these are followed by a reference key from the reference_files mapping in the top level of the config file, and are parsed at run time into the inputs for the workflow (i.e. use of "stage-cnv_generate_bed_vep.gene_panels": "INPUT-genepanels" would result be replace by project-Fkb6Gkj433GVVvj73J7x8KbV:file-GVx0vkQ433Gvq63k1Kj4Y562, correctly formatted as a $dnanexus_link mapping)

  Currently supported placeholder strings include:

  • INPUT-clinical_indications : ';' separated string of clinical indications
  • INPUT-panels : ';' separated string of panels
  • INPUT-test_codes : '&&' separated string of test codes
  • INPUT-sample_name : string of sample name from manifest

  These are added to the config via utils.add_dynamic_inputs from kwargs generated at run time specified here.

---

What does this app output

• summary_report (file) - text summary file with details on jobs run and any samples / tests excluded from analysis

---